A new pattern locating algorithm is presented for high precision calibration of the machine vision sensor.This locating algorithm includes accurate target detection and pattern matching.The developed subpixel estimation method based on moment is immune to the pattern shape
so it overcomes the non-circular disadvantage of circular target patterns by photography transformation
and ensures the detection accuracy.The proposed matching algorithm can adaptively adjust the matching direction and distance metrics according to the information of neighbor targets.This matching algorithm is stable
automated for any attitude of the pattern image.The new locating algorithm is used to recognize the new design pattern
the experiments show the locating algorithm has high precision
Least-squares fitting of circles and ellipses [J] . Walter Gander,Gene H. Golub,Rolf Strebel. BIT . 1994 (4)
A formal definition of the Hough transform: Properties and relationships [J] . J. Princen,J. Illingworth,J. Kittler. Journal of Mathematical Imaging and Vision . 1992 (2)